Estimation of exogenous drivers to predict COVID-19 pandemic using a method from nonlinear control theory

Abstract

The currently ongoing COVID-19 pandemic confronts governments and their health systems with great challenges for disease management. Epidemiological models play a crucial role, thereby assisting policymakers to predict the future course of infections and hospitalizations. One difficulty with current models is the existence of exogenous and unmeasurable variables and their significant effect on the infection dynamics. In this paper, we show how a method from nonlinear control theory can complement common compartmental epidemiological models. As a result, one can estimate and predict these exogenous variables requiring the reported infection cases as the only data source. The method allows to investigate how the estimates of exogenous variables are influenced by non-pharmaceutical interventions and how imminent epidemic waves could already be predicted at an early stage. In this way, the concept can serve as an “epidemometer” and guide the optimal timing of interventions. Analyses of the COVID-19 epidemic in various countries demonstrate the feasibility and potential of the proposed approach. The generic character of the method allows for straightforward extension to different epidemiological models.

An Analytical Framework for Understanding Infection Progression Under Social Mitigation Measures

“…Refs. [32][33][34][35][36]), i.e., only the first phase of epidemics (and the social measure introduction) is considered. We implement our model deterministically, as publicly available COVID-19 counts [63] are very high in most countries, making the relative importance of fluctuations low, and the deterministic description appropriate [62].…”

“…The model is a generalization of the compartmental SEIR model -socalled SPEIRD [57,58] -that accounts for the influence of social measures by mathematically representing them as an effect that removes susceptibles from the transmission process (and places them in a newly introduced Protected compartment). A similar approach to introduce the effects of distancing in compartmental models can be found in [32][33][34][35][36], and is in certain aspects advantageous over simply varying transmission coefficients in time, implemented in many models [4,8,22,[37][38][39]. SPEIRD also introduces additional compartments that directly correspond to epidemiological observables: the number of detected cases and the number of fatalities.…”

“…( 48) in Eq. (34), and by taking into account Eq. ( 29), we obtain the expression (16) for the general solution of Eq.…”

“…In particular, the attempts to combat the disease led to the introduction of social distancing measures on the scales hardly imaginable before [24]. While it is evident that they have a serious impact on the epidemiological dynamic, their effects are not yet well understood [25][26][27], despite a lot of undergoing effort to incorporate the influence of these measures into the existing epidemiological models [4,8,22,[28][29][30][31][32][33][34][35][36][37][38][39]. Common for all these extensions of epidemiological models (both those related to the effects of mitigation measures and to the inclusion of other relevant factors), is rely-…”

Noise Resistance of Next Generation Reservoir Computing: A Comparative Study with High-Order Correlation Computation

“…Analysis and prediction of data play an important role in people's production and life, such as weather prediction, environmental pollution control, earthquake prediction, financial data analysis, speech recognition, image processing, and aircraft control [1][2][3][4][5][6][7][8][9][10]. For the field of time-series prediction, various methods have been proposed and obtained more satisfactory prediction results, such as system dynamics reconstruction, neural network prediction, etc.…”